12398 • The Journal of Neuroscience, August 31, 2011 • 31(35):12398–12412

Behavioral/Systems/Cognitive Equiluminance Cells in Visual Cortical Area V4

Brittany N. Bushnell,1 Philip J. Harding,1 Yoshito Kosai,1 Wyeth Bair,1,2 and Anitha Pasupathy1 1Department of Biological Structure and Washington National Primate Research Center, University of Washington, Seattle, Washington 98195, and 2Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford OX1 3PT, United Kingdom

We report a novel class of V4 neuron in the macaque monkey that responds selectively to equiluminant colored form. These “equilumi- nance” cells stand apart because they violate the well established trend throughout the visual system that responses are minimal at low luminance contrast and grow and saturate as contrast increases. Equiluminance cells, which compose ϳ22% of V4, exhibit the opposite behavior: responses are greatest near zero contrast and decrease as contrast increases. While equiluminance cells respond preferentially to equiluminant colored stimuli, strong hue tuning is not their distinguishing feature—some equiluminance cells do exhibit strong unimodal hue tuning, but many show little or no tuning for hue. We find that equiluminance cells are color and shape selective to a degree comparable with other classes of V4 cells with more conventional contrast response functions. Those more conventional cells respond equally well to achromatic luminance and equiluminant color stimuli, analogous to color luminance cells described in V1. The existence of equiluminance cells, which have not been reported in V1 or V2, suggests that chromatically defined boundaries and shapes are given special status in V4 and raises the possibility that form at equiluminance and form at higher contrasts are processed in separate channels in V4.

Introduction nance contrast on chromatic responses is typically facilitatory. Responses of neurons in primate area V4, an intermediate Most neurons in V1, V2, and V3 are sensitive to both luminance stage along the ventral visual pathway, are sensitive to the and color (Gouras and Kruger, 1979; Thorell et al., 1984; Hubel color (Zeki, 1973; Schein and Desimone, 1990) and the ach- and Livingstone, 1990; Lennie et al., 1990; Johnson et al., 2001; romatic luminance of visual stimuli (Schein and Desimone, Hansen and Gegenfurtner, 2007), and the contrast response 1990). But, because studies seldom vary chromaticity and lu- functions in these areas show a characteristic V-shaped pattern minance simultaneously, we do not yet know how luminance (i.e., responses increase, and saturate, with increasing luminance and chromatic signals are multiplexed in area V4. This is an contrasts) (Sclar et al., 1990; Albrecht, 1995; Gegenfurtner et al., important question to address because (1) V4 is known to play an 1997; Kiper et al., 1997). A small fraction of V1 neurons respond important role in form processing (Desimone and Schein, 1987; more strongly to equiluminant stimuli than to luminance stim- Kobatake and Tanaka, 1994); (2) in natural vision, form can be uli, but because these neurons are typically low pass and unori- defined by chromatic contrast, luminance contrast, or both ented (Lennie et al., 1990; Johnson et al., 2001), they are unlikely (Hansen and Gegenfurtner, 2009); and (3) psychophysical stud- to contribute to form encoding. The V1 and V2 neurons ies suggest that both chromatic and luminance contrasts contrib- equipped to encode object boundaries multiplex luminance and ute to form perception (Cavanagh, 1991). Deciphering how color signals (Lennie et al., 1990; Johnson et al., 2001; Shapley and luminance and chrominance signals are combined across the V4 Hawken, 2002; Hansen and Gegenfurtner, 2007). population will advance our understanding of how cortex pro- To discover how luminance contrast modulates the responses duces a unified, invariant perception of form. of V4 neurons to colored stimuli, we studied the responses of Evidence from early stages of visual processing suggests that single neurons in awake monkeys to a set of 25 chromaticities chrominance-defined and luminance-defined form information presented at four different luminance contrasts. As in V1, a ma- is carried by the same neurons and that the influence of lumi- jority of V4 neurons responded to both chromatic and achro- matic stimuli, and luminance contrast had a facilitatory influence on chromatic responses. For a sizable minority, however, equilu- Received April 14, 2011; revised June 25, 2011; accepted July 8, 2011. minant color stimuli evoked the strongest responses, which de- Author contributions: P.J.H., W.B., and A.P. designed research; B.N.B., P.J.H., Y.K., and A.P. performed research; creased when luminance contrasts were added to the chromatic A.P. analyzed data; W.B. and A.P. wrote the paper. stimuli. These equiluminance cells do not fall into previously This work was supported by National Eye Institute Grant R01 EY018839, The Whitehall Foundation, University of Washington Vision Core Grant P30 EY01730, and National Center for Research Resources Grant RR00166. W.B. is established cell classes in V1 or V2, and their shape-selective supportedbytheWellcomeTrustandSt.John’sCollege,Oxford(Oxford,UK).WethankYasmineEl-Shamayleh,Greg properties suggest that these neurons can contribute to the en- Horwitz, and Raghu Pasupathy for helpful discussions and comments. Jalal Baruni and National Primate Research coding of equiluminant colored form. Center Bioengineering provided technical support. Correspondence should be addressed to Anitha Pasupathy, Department of Biological Structure and Washington NationalPrimateResearchCenter,UniversityofWashington,1959PacificStreetNortheast,HSBG-520,Universityof Materials and Methods Washington Mailbox 357420, Seattle, WA 98195. E-mail: [email protected]. Surgery and animal behavior DOI:10.1523/JNEUROSCI.1890-11.2011 Two rhesus monkeys (Macaca mulatta; one male and one female) were Copyright © 2011 the authors 0270-6474/11/3112398-15$15.00/0 surgically implanted with custom-built head posts attached to the skull Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12399 with orthopedic screws. After fixation training (see below), a recording chamber was implanted based on structural MRI scans. A craniotomy was performed in a subsequent surgery. For detailed surgical procedures, see Bushnell et al. (2011). All animal procedures conformed to NIH guidelines and were approved by the Institutional Animal Care and Use Committee at the University of Washington. Animals were seated in front of a computer monitor at a distance of 57 cm and were trained to fixate a 0.1° white dot within 0.5–0.75° of visual angle. Eye position was monitored using a 1 kHz infrared eye-tracking system (Eyelink 1000; SR Research). Stimulus presentation and animal behavior were controlled by Linux-based custom software (PYPE; orig- inally developed in the Gallant Laboratory, University of California, Berkeley, Berkeley, CA). Each trial began with the presentation of a fix- ation spot at the center of the screen. Once fixation was acquired, four to six stimuli were presented in succession, each for 300 ms (500 ms for 22 cells), separated by interstimulus intervals of 200 ms. Stimulus onset and offset times were based on photodiode detection of synchronized pulses in the lower left corner of the monitor.

Data collection During each recording session, a single dura-puncturing microelectrode (FHC), 250 ␮m in diameter, was lowered into the cortex using an eight- channel acute microdrive system (Gray Matter Research). Signals from the electrode were amplified and filtered and single-neuron activity was isolated using a spike sorting system (Plexon Systems). We targeted dorsal V4 and our electrode penetrations were largely in the prelunate gyrus and occasionally in the adjoining bank of the lunate Figure1. Colorandshapestimuli.A,The25chromaticitiesusedinthestudyareshownintheCIE sulcus. Previous studies suggest that color selective units are clustered in chromaticitydiagram.Thecolorsoccupiedthreetrianglesatincreasingdistancesfromtheachromatic patches, recently termed “globs” (Conway et al., 2007). While the precise point(no.1)andrepresentedthreelevelsofcolorcontrast:colors2–7composedthelowcolorcontrast location of globs varies across animals, combining fMRI and physiology, band; 8–13, the mid color contrast band; and 14–25, the high color contrast band. Each of the 25 Conway and colleagues suggested that globs were less prevalent in dorsal chromaticities was presented at four different luminance contrasts relative to the achromatic V4. We do not know the precise location of globs in our animals, but background labeled 1. B, Shape stimuli used to characterize the shape preferences of neurons. 15–20% of our recorded neurons had hue tuning (see Results) that was A subset (10–29) of these shapes was presented in eight orientations at 45° intervals to study roughly consistent with that attributed to glob regions. shape selectivity. Some shapes that show rotational symmetry were presented at fewer rota- tions. For example, shapes labeled 1 and 2 were presented at two and four rotations, respec- Visual stimuli tively. Shape labeled 1 was used to characterize color and luminance preferences of example Visual stimuli were presented on a CRT monitor (40.6 ϫ 30.5 cm; 97 Hz ϫ neuronswhosedataareshowninFigure2,A,B,andD;shapelabeled2wasusedtocharacterize frame rate; 1600 1200 pixels) calibrated with a spectrophotoradiom- the example neuron whose data are shown in Figure 2C. eter (PR650; PhotoResearch). Stimuli were presented against an achro- matic gray background (Fig. 1A, symbol 1) of mean luminance of 5.4 cd/m 2. For each isolated unit, an initial qualitative preferred stimulus (shape, color, orientation) and a rough receptive field (RF) location were point of the animal or the neuron under study. So, for a subset of neurons identified using a variety of shapes under the experimenter’s control. that responded strongest to 0% contrast stimuli, we sampled the lumi- This was followed by an automated RF mapping procedure that pre- ϭ Ϯ Ϯ nance axis more finely, characterizing responses at Clum 0, 2.5, 5, sented the initial preferred stimulus in a densely sampled grid; the refined Ϯ25, and Ϯ50%. RF center was based on a two-dimensional Gaussian fit to the data. To characterize shape selectivity, for 118 of the 202 neurons, we stud- To characterize tuning for chromaticity and luminance contrast, we ied the responses to 10–29 shapes presented in eight orientations at 45° studied the responses to a preferred shape presented in 25 chromaticities intervals in a preferred color based on the color ϫ luminance test above. 2 each at four different luminances (2.7, 5.4, 8.1, and 12.1 cd/m ). The Some shapes were tested only at two or four rotations due to rotational preferred shape was defined as that which evoked the best response dur- symmetry. The tested shapes (Fig. 1B) were a subset of those previously ing an initial characterization performed under manual control (varying used to study V4 (Pasupathy and Connor, 2001; Bushnell et al., 2011) and position, orientation, and size) with the following: drifting and flashing are known to be effective at driving responses of a large fraction of V4 bars, ellipses, and a set of shapes (Fig. 1B) that were designed to explore neurons. a range of convex and concave contour features in the context of a study on partial occlusion (Bushnell et al., 2011). The preferred shape was sized Data analysis such that all parts of the shape were entirely within the RF of the cell All results presented here are based on mean responses obtained by av- under study. The four luminances tested corresponded to (Weber’s) eraging the firing rate in the interval between stimulus onset and stimulus Ϫ contrasts (Clum)of 50, 0, 50, and 125%, respectively. The tested colors offset across stimulus repetitions. For each cell, we also computed a spanned the full gamut of the monitor and sampled three triangles at latency-adjusted average rate by counting spikes in a window equal to increasing distances from the achromatic point (Fig. 1A, symbol 1) in the stimulus duration offset by the response latency of the cell at the corre- CIE color space. Colors 2–7 sampled the innermost triangle and the sponding luminance contrast. Results based on these latency-adjusted lowest color contrast band; colors 8–13 sampled the middle triangle and average rates were similar to those reported here. Color and luminance mid-color contrast band; the remaining colors, which lie on the outer- characterizations were based on six pseudorandom repeats, shape results most triangle sampled the highest color contrast band in our study. CIE were based on five repeats, and fine-scale sampling along the contrast axis xy coordinates for the background and the vertices of the outermost (see Fig. 8) was based on 20 repeats. Responses to the achromatic stim- triangle (nos. 16, 24, and 20) are as follows: background ϭ (0.33, 0.33), ulus at 0% contrast (blank gray screen) provided a measure of the base- red ϭ (0.5, 0.32), green ϭ (0.28, 0.45), and blue ϭ (0.17, 0.09). In this line firing rate. Results presented here are based on mean responses experiment, the equiluminance point (0% contrast) corresponds to pho- without subtraction of baseline firing, but results were similar with base- tometric equiluminance, which may not match the true equiluminance line subtraction. 12400 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4

Contrast response functions. To construct contrast response functions with which the neuron combines these cone signals (Lennie et al., 1990);

(see Fig. 2, bottom row), for each cell, we first identified colors that and Rb, the baseline firing rate. To obtain the cone contrasts (CL, CM, CS) evoked responses significantly different from baseline (randomization t for the various stimuli, the CIE coordinates were first converted to cone test, p Ͻ 0.05). Then, for each chromaticity that evoked significant re- excitations based on Smith and Pokorny (1975) fundamentals (Cole and sponses at one or more luminance contrasts, responses to the different Hine, 1992). Then the cone contrast, C, for each of the cone signals was ϭ Ϫ Ϫ contrasts were normalized to the range 0–1 [Rnorm R Rmin/(Rmax given by the following: Rmin)]. The within-chromaticity-normalization ensured that the con- ϭ ͑ Ϫ ͒ trast response functions were not dominated by the colors that evoked C Estim Ebg /Ebg, (2) the strongest responses. The shape of the contrast response functions was characterized by two sets of parameters: rbright and Sbright, which mea- where Estim and Ebg represent the cone excitation for the stimulus and sured the linear correlation coefficient and slope, respectively, between background, respectively. Because some cells responded only to positive Ն Clum 0 and the normalized responses, and rdark and Sdark, which mea- or negative contrasts, whereas others responded to both, we imple- Յ sured the same between Clum 0 and the normalized responses. These mented half-wave and full-wave rectified versions of the above model bright (rbright and Sbright) and dark (rdark and Sdark) parameters were (Lennie et al., 1990). For each cell, relative cone weights, WL, WM, and ϫ ϫ based on 3 n and 2 n data points, respectively, where n represents the WS, were obtained by dividing the corresponding absolute cone weights number of chromaticities that evoked responses significantly different (wL, wM, wS from Eq. 1 above) by the sum of the magnitude of the three from baseline for the neuron in question. Mean contrast response func- weights (Lennie et al., 1990). tions in Figures 2 and 4 were constructed by averaging the normalized Shape selectivity. Shape selectivity was characterized with two mea- responses across chromaticities. sures—the fraction of shapes that evoked a significantly different re- Ͻ Separability of the influence of chromaticity and luminance contrast on sponse from baseline (Fshape_sig; randomization t test, p 0.05) and the V4 responses. The validity of pooling contrast response functions across fraction that evoked greater than one-half of the maximum response chromaticities, as described above, depends on the separability between (Fshape_hmax). For shape selectivity, we used the fraction rather than num- the influences of color and luminance contrast on single neuron re- ber because different neurons were tested with different numbers of sponses. To directly assess the extent of separability, we used singular shapes (see Materials and Methods). value decomposition (SVD), which has been previously used for the same purpose in vision (Mazer et al., 2002) and audition (Depireux et al., ϫ Results 2001; Pen˜a and Konishi, 2001). This method decomposes the color Equiluminance cells in V4 luminance contrast response matrix, R, of each neuron into the form To examine how luminance contrast modulates neuronal re- UTSV. U and V are a set of orthogonal vectors and S is a diagonal matrix of singular values; when these are combined, R is fully reconstructed. If sponses to colored shape stimuli in visual area V4, we studied the the influence of color and luminance are completely separable, only responses of 202 neurons to a preferred shape presented at 25 the first diagonal term in S will be nonzero. For every cell, we quan- chromaticities at each of four different luminance contrasts. Be- tified the singular matrix S and computed a separability index, SI, given cause nearly all neurons in the LGN, V1, V2, V3, V4, and V5/MT ϭ ␣ 2 ⌺ ␣ 2 by the relative magnitude of the first singular value: SI (1) / i (i) , show responses that increase, or increase and saturate, as a func- where ␣(i)istheith diagonal term of S. SI values range from 0 for tion of luminance contrast (Sclar et al., 1990; Albrecht, 1995; nonseparable matrices to 1 for completely separable matrices. Across our Gegenfurtner et al., 1997; Kiper et al., 1997; Reynolds et al., 2000; population of V4 neurons, SI was Ͼ0.77 for all except one cell, and Ϯ Ϯ Lee et al., 2007), we expected to find many V4 neurons that in- mean SE was 0.93 0.054. This indicates that the influence of color creased their responses with luminance contrast, even for colored and luminance contrast on V4 responses were largely separable. For ϫ stimuli. In keeping with this expectation, a majority of V4 cells comparison, in V1 the mean SI for spatial frequency orientation was ϳ 0.9 Ϯ 0.09 (Mazer et al., 2002). ( 64%) in our database showed responses to colored stimuli that Peristimulus time histograms. To construct population peristimulus increased in magnitude and decreased in latency with increasing time histograms (PSTHs), we first constructed single cell PSTHs at every luminance contrast. Using example neurons, we first describe luminance contrast by averaging and smoothing (Gaussian, ␴ ϭ 5 ms) three classes of such cells (Fig. 2A–C), and then introduce a responses across chromaticities. These were normalized by the peak re- fourth class that exhibited the opposite behavior (Fig. 2D). sponse across all contrasts and then averaged across cells. Figure 2A demonstrates a “bright” cell, which responded best Color selectivity. Color selectivity was quantified for each cell in several when its preferred pattern (a star rotated by 15°, inset) was ways. We used two simple metrics—the number of chromaticities that brighter than the background. This is evident in the raster plots evoked a response significantly different from baseline (randomization t (top panels), in which raw responses are grouped by luminance test, p Ͻ 0.05) at one or more luminance contrasts (N ) and the col_sig contrast. Within each panel, trials are further grouped by color, number of chromaticities that evoked greater than one-half of the max- but this is not apparent because color tuning (analyzed below) imum response (Ncol_hmax) at the luminance contrast that evoked the best response. We also constructed hue tuning curves at each of the three was broad for this neuron. The purpose of the raster plots is to color contrast bands (Fig. 1A) by representing the neuronal response as illustrate the consistent preference for the positive nonzero con- a function of direction in CIE space. We then assessed whether hue trast across colors. The PSTHs below the rasters summarize the tuning was significantly different from uniform tuning using the Ray- average response time course and show that stimuli brighter than the Ͻ ϭ leigh test of circular uniformity (p 0.05, Bonferroni corrected). Fol- background (Clum 50 and 125%) evoked stronger responses than lowing Conway et al. (2007), strength of unimodal tuning was quantified ϭ ϭϪ those equiluminant to (Clum 0%) or darker (Clum 50%) than as the resultant vector length (i.e., the weighted average of the color the background. In addition, the response onset latency decreased direction vectors and the corresponding responses). To determine progressively with increasing C (see PSTH inset), consistent with whether color responses of V4 neurons can be modeled as a linear func- lum previous reports (Gawne et al., 1996; Maunsell et al., 1999). The tion of cone excitation, for every neuron, we estimated the cone weights, luminance contrast tuning function (Fig. 2A, bottom panel) (see wL, wM, wS, that provided the best fit for the observed responses per the following equation: Materials and Methods) was characterized by two parameters, rdark and rbright, which are the correlation coefficients between ϭ ϫ ϩ ϫ ϩ ϫ ϩ Յ Ն R wL CL wM CM wS CS Rb, (1) neuronal responses and Clum 0orClum 0, respectively. For this cell, rbright and rdark were 0.89 and 0.26, respectively. Figure where R represents the neuronal response; C , C , and C , the cone 2B illustrates a “dark” cell, which responded preferentially to L M S ϭ contrast signal relative to the background; wL, wM, and wS, the weights stimuli (a star at 45°) at negative luminance contrasts (Clum Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12401

A Bright B Dark C Contrast D Equiluminance

-50

0

50 Luminance contrast (%)

125

60 40 30 −50 30 0 50 40 20 20 125 20 40 ms 10 10 20 Spikes/sec 0 0 0 0 0 200 400 0 200 400 0200400 0200400 Time relative to stimulus onset (msec) 1 1 1 1

0.5 0.5 0.5 0.5

0 0 0 0 −50 0 50 100 −50 0 50 100 −50 0 50 100 −50 0 50 100 Contrast (%) Normalized response

Figure 2. Single neuronal examples of four types of luminance contrast selectivity in V4. A–D, Raster plots show single trial responses for six repeats of 25 colors grouped by luminance contrast (rows).RastersareorderedbycolornumberinginFigure1A.Thus,withineachrasterplot,rows1–6fromthebottomdepictresponsestocolor1intheCIEdiagram,7–12representresponsestocolor 2, etc. The x-axis (for rasters and PSTHs) represents time relative to stimulus onset. Stimuli were presented for 300 ms. PSTHs were Gaussian-smoothed with ␴ ϭ 5 ms; line color represents luminance contrast. The inset in A shows enlarged version of the 40 ms epoch starting 30 ms after stimulus onset to facilitate response latency comparison across luminance contrasts. Contrast response functions (bottom panel) show mean responses, normalized across colors (see Materials and Methods), versus luminance contrast. Error bars indicate SEM. Example neurons responded strongest to positive luminance contrast (bright cell) (A); negative luminance contrast (dark cell) (B); high contrast, either positive or negative (contrast cell) (C); and stimuli that were nominally equiluminant with the background (equiluminance cell) (D).

Ϫ50%). Stimuli at 0, 50, and 125% contrasts evoked weak re- evoked the strongest responses, while stimuli at higher contrasts sponses. We call this neuron a dark rather than an “off” cell (Ϫ50, 50, and 125%) evoked much weaker responses. Unlike because, while stimuli darker than the background evoked strong results from analogous experiments in V1 (Gouras and Kruger, responses, offset of bright stimuli did not (see responses to 1979; Thorell et al., 1984; Hubel and Livingstone, 1990) and V2 ϭ Ϫ Clum 125%). For this neuron, rbright and rdark were 0.2 and (Kiper et al., 1997), the responses of this neuron were suppressed Ϫ0.98, respectively. Figure 2C shows responses of a “contrast” when luminance contrast was added to a patch of color that ap- cell studied with a bar oriented at 60° relative to the vertical. This proximated background luminance. The resulting contrast re- ϭϪ neuron responded equally to stimuli at positive and negative con- sponse function has an inverted V shape with rbright 0.82 and ϭ ϭ trasts; stimuli at Clum 0% alone evoked weak responses. The rdark 0.99. Even though responses are strongest for 0% contrast contrast tuning function of this neuron was the classical V shape stimuli, the latency at 0% is longest (red trace), just as in the ϭ ϭϪ with rbright 0.77 and rdark 0.91. previous examples. In other words, despite the weaker amplitude, In striking contrast to these examples, Figure 2D shows data responses to high contrast stimuli emerge earlier than responses from an “equiluminance” cell for which stimuli at 0% contrast to low contrast stimuli (Fig. 2D, compare red, blue, and black 12402 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4

A Bright B Dark C Contrast D Equiluminance

-50% %) 0% lum

50% Luminance contrast (C

0.6 18 24 27 30 Spikes/sec

125% 0.3 9 12 15 15 CIE y

0.0 0 0 2 0 0.0 0.3 0.6 CIE x

Figure3. Colorpreferencesandconsistencyofluminancecontrastpreferenceacrosschromaticities.MeanresponsesoftheexampleneuronsinFigure2werelinearlyinterpolatedtoconstructthe responsesurfacesshownhereasafunctionofchromaticityandluminancecontrast.Eachcolumndepictsanexampleneuron,andeachrowrepresentsoneofthetestedluminancecontrasts.Foreach panel, the CIE x- and CIE y-coordinates are plotted along the x- and y-axes, and the color scale denotes the strength of the mean response evoked by each stimulus. Mean responses were calculated from stimulus onset to stimulus offset across six presentations. Color scale runs from pale blue (weak responses), through black (intermediate responses), to bright red (strong responses). A, Bright ϭ cell. Preference for positive luminance contrasts is clearly evident, but color preference is not: most chromaticities at Clum 125% (bottom row) evoked strong responses (red) from this neuron. B, ϭϪ Dark cell. This neuron exhibited moderate color tuning with preferred responses clustered close to the red region of the CIE diagram at Clum 50%. Preference for negative luminance contrasts ϭ is consistent across chromaticities. C, Contrast cell. There is some clustering of preferred responses (red points) close to the achromatic point; responses at Clum 0 were consistently weak at all ϭ chromaticities.D,Equiluminancecell.Moderateclusteringofpreferredresponsesalongthegreen-magentaaxisatClum 0;responsesneartheachromaticpoint,whichrepresentsthebackground ϭ color, were weak; this results in an annulus clustering of preferred responses. Responses of all chromaticities, except the background color, were strongest at Clum 0.

ϭ PSTHs). This violates the frequently observed relationship be- sponses significantly different from baseline, Ncol_sig 25) (see tween response latency and magnitude, but it supports, in strik- Materials and Methods) and 88% of colors evoked greater than ing fashion, past observations of a dissociation between latency half-maximum response (Ncol_hmax, number of colors that evoked and magnitude in favor of a relationship between latency and greater than half-maximum response ϭ 22). Furthermore, a prefer- luminance contrast (Gawne et al., 1996; Carandini et al., 1997; ence for positive luminance contrast is clearly evident and consistent ϭ Maunsell et al., 1999). Put simply, at high luminance contrast, across chromaticities—responses were strongest at Clum 125% for ϭ neuronal responses are fast, even if they are weak because of a all except one chromaticity [no. 6 (Fig. 1A) peaks at Clum 50%] ϭϪ nonoptimal stimulus form [nonpreferred orientation, as in the and responses were weaker at Clum 50 and 0% for all study by Gawne et al. (1996)] or a nonoptimal contrast, including chromaticities. ϭ high contrasts, for equiluminance cells here. The dark cell (Fig. 3B), which responded strongest at Clum Ϫ ϭ ϭ The contrast response functions in Figure 2 were constructed 50%, exhibited broad color selectivity (Ncol_sig 23; Ncol_hmax by pooling across chromaticities (see Materials and Methods), 15) with preferred chromaticities clustered toward red (no. 16) in the but this would be reasonable only to the extent that the contrast CIE space. The effect of luminance contrast was again highly consis- ϭϪ response curves remained qualitatively consistent across chro- tent: responses were strongest at Clum 50% for all except chro- maticity. To assess consistency for our example cells, and to visu- maticity no. 12. The contrast cell (Fig. 3C) responded strongly to all alize their color tuning, the responses from the rasters in Figure 2 nonzero contrasts, and all colors evoked significant responses at one are replotted in Figure 3 as surfaces in CIE space for each lumi- or more of those contrasts. Color tuning was broad (Ncol_hmax was 18 ϭϪ ϭ nance contrast. The bright cell (Fig. 3A) showed weak color se- at Clum 50% and 16 at Clum 50 and 125%), and the preferred ϭ lectivity: at Clum 125%, all tested chromaticities evoked responses appeared to be clustered in CIE space. All except one of the significant responses (the number of colors that evoked re- chromaticities (no. 11) exhibited V-shaped contrast response func- Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12403

Equiminance cells: Bright cells: 1 1 n = 12 n = 12 rbright < 0 rdark > 0

0.5 0.5

rbright ~ 0 rbright > 0 n = 44 rdark > 0 rdark > 0 Normalized response Normalized response 0 0 −50 0 100 −50 0 100 1 n = 25 Contrast % 1

Flat cells: 0.5 rbright & rdark ~ 0 0.5 rbright > 0 Peak cells: r ~ 0 r & r ~ 0 and dark bright dark 0 peak/trough of contrast −50 0 100 response function at Clum = 50 0 Contrast % dark r

1 n = 16

−0.5 rbright < 0 0.5 rdark < 0

−1 −1 −0.5 0 0.5 1 0 Contrast cells: −50 0 100 1 rbright 1 n = 11 n = 51 rbright ~ 0 rdark < 0 Dark cells: 0.5 0.5

Normalized response rbright > 0 rdark < 0 0 0 −50 0 100 Normalized response −50 0 100 Contrast % Contrast %

Figure4. Classificationofcellsbasedoncontrastresponsefunctions.Cellswereclassifiedintooneofsixcategories(bright,dark,contrast,equiluminance,flat,andpeak)basedontwocorrelation Ն values: rbright (x-axis of scatterplot), which measures the coefficient of correlation between Clum and neuronal responses to the different chromaticities for Clum 0, and rdark ( y-axis of scatterplot), Յ which measures the correlation between Clum and neuronal responses for Clum 0. Only those chromaticities that evoked responses significantly greater than baseline at one or more luminance contrasts were included in the rbright and rdark calculations (see Materials and Methods). Mean contrast response functions grouped according to the properties of rbright and rdark (listed in the corresponding breakouts; for details, see Results) for bright, dark, contrast, and equiluminance cells are shown in the top right, bottom left, bottom right, and top left respectively. ϳ denotes not significantly different from 0. Number of cells in each breakout panel is indicated by n.

ϭ tions with minimum response at Clum 0%. Finally, the equilumi- separability index using singular value decomposition (see Mate- ϭ Ͼ nance cell responded best to chromaticities at Clum 0%. Color rials and Methods) and found this index to be 0.96 for all four ϭ ϭ tuning for this cell was also broad (Ncol_sig 22; Ncol_hmax 13). cells, and 0.93 on average across our database. This confirms that Preferred chromaticities ranged from green (no. 24) to magenta (no. the influence of color and luminance contrast were largely sepa- 18) with weak responses close to the achromatic point, resulting in rable. Therefore, we will next average across chromaticities to an annulus-shaped clustering of preferred responses. All chromatici- characterize the luminance contrast behavior of our cells. We ties except the achromatic point, which is the background color at return later to summarize color tuning in more detail at the end. ϭ Clum 0% (and thus no stimulus), exhibited strongest responses at ϭ Clum 0%. Population results In summary, these four neurons showed qualitatively consistent Figure 4 summarizes the relationship between luminance con- trends with luminance contrast across a broad range of chromatici- trast and neuronal response for 197 of the 202 neurons for which ties. To verify this quantitatively, we computed a color luminance at least two colors evoked responses significantly different from 12404 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4 baseline (randomization t test, p Ͻ 0.05). Based on the shape of A the contrast response functions and the extracted parameters 0.02 Bright rbright and rdark, cells were classified into one of six categories Dark (indicated by colored symbols in Fig. 4). Bright cells (green dots, Contrast Ͻ 49 of 197; Fig. 2A) responded weakly to Clum 0 and strongly to Equiluminance Ͼ ϭ Clum 0. Responses at Clum 0 varied across neurons and were Flat Ͻ 0.01 either weak and comparable with responses at Clum 0, strong Peak Ͼ and comparable with responses at Clum 0, or intermediate. This resulted in contrast response functions with three characteristic Ͼ patterns (Fig. 4, breakouts in the top right): (1) rbright 0, rdark not significantly different from zero (points along the positive 0 Ͼ dark x-axis, main panel); (2) rdark 0, rbright not significantly different S from zero (points along the positive y-axis); and (3) rbright and Ͼ rdark 0 (points in the first quadrant). Dark cells (blue dots, 27 of 197; Fig. 2B) responded strongest to stimuli that were darker Ͻ Ͼ −0.01 than the background (Clum 0) and weakly to Clum 0. At ϭ Clum 0, responses were either weak and comparable with re- Ͼ Ͻ sponses at Clum 0(rdark 0 and rbright not significantly different from 0; points along the negative y-axis), or intermediate be- Ͼ Ͻ Ͻ −0.02 tween responses to Clum 0 and Clum 0(rbright and rdark 0, points in the third quadrant). Contrast cells (black dots, 51 of −0.01 00.01 S 197), had a V-shaped response profile and responded best to bright high-contrast stimuli of either polarity. These cells occupied the B Ͼ Ͻ fourth quadrant (rbright 0 and rdark 0). Equiluminance cells Background lum. ϭ 1 (cyan dots, 44 of 197) responded best to stimuli at Clum 0; 2.7 stimuli brighter or darker than the background evoked weaker 5.4 responses. Thus, equiluminance cells had inverted V-shaped 8.1 Ͻ Ͼ contrast response functions (rbright 0 and rdark 0) and occu- pied the second quadrant. For the remaining two groups, “flat” 0.5 (red dots, 21 of 197) and “peak” cells (magenta dots, 5 of 197), rbright and rdark were not significantly different from zero. Peak cells, however, showed significant modulation of responses with ϭ Normalized response stimulus contrast and had a peak or trough at Clum 50%. These are consistent with peak-shaped luminance response functions in 0 V1 and V2 reported by Peng and Van Essen (2005). The catego- 2.7 5.4 8.1 12.1 ries depicted here are consistent with those obtained using Stimulus luminance (cd/m2) K-means clustering with number of clusters set to 5 or 6 (data not shown). Results were also similar when we used the contrast re- Figure 5. Slopes of the contrast response functions and example cell studied at three back- sponse function associated with the chromaticity that evoked ground luminances. A, Each dot represents a cell. Sbright (x-axis of scatterplot) represents the slope given by a linear regression fit between C and neuronal responses to the different strongest responses as the basis for our categorization. lum chromaticities for C Ն 0, and S ( y-axis of scatterplot) represents the slope between C The parameters r and r provide a measure of the consis- lum dark lum bright dark and neuronal responses for C Յ 0. The dot colors and analysis details are as in Figure 4. tency of contrast response curves across chromaticities but not of the lum Because the positive contrast range for Clum was 0–125% and the negative contrast range was magnitude of change in response as a function of luminance con- 0toϪ50% and responses were normalized to lie between 0 and 1, the maximum slope limit trast. Thus, we examined the slopes, Sbright and Sdark, between nor- along the x- and y-axes are 0.008 (1/125) and 0.02 (1/50) respectively. Across our population, Յ Ն malized responses and Clum 0orClum 0, respectively. Figure 5A slopesspanthisentirerange.B,Normalizedresponses( y-axis)plottedasafunctionofstimulus shows the relationship between the slopes, Sbright and Sdark, for the luminance (x-axis) for an example equiluminance cell studied at three different background different categories of cells. The spread of slopes is consistent with luminances. At each background luminance (denoted by line style), individual chromatic con- the spread of correlation coefficients in Figure 4. trast response functions were constructed as detailed in Materials and Methods; their mean While the precise boundaries between categories are not crit- (ϮSEM) is plotted here as a function of stimulus luminance rather than contrast. Responses ical, it is important to note several characteristics of the spread of peaked at the stimulus luminance that matched the background luminance, suggesting that points in Figures 4 and 5A. First, equiluminance cells (cyan) rep- luminance contrast, rather than absolute luminance, dictated the response. resent a sizable proportion across the population. Results were comparable in both animals (equiluminance cells: monkey M, binations of Sbright and Sdark are instantiated more often than 21%; monkey C, 24%). When the stimuli were presented on a others. To further test whether equiluminance cells are the result lower background luminance (2.7 cd/m 2), we also found a simi- of random variation in the peak position of the contrast response lar proportion of equiluminance cells (16 of 94 Ϸ 17%) across a function, we quantified the number of cells with peaks at each of mostly different set of 94 cells (except 6 cells tested at both back- the four tested luminance values. The position of the peak in the ground luminances). Second, there is a continuum from equilu- mean contrast response function was not evenly distributed be- minance cells to contrast cells that passes through the origin (flat tween the four tested luminance contrasts (␹ 2 test, p Ͻ 0.001). ϭ cells). Third, the spread across all points is not radially symmet- Peaks at Clum 50%, the intermediate positive contrast, were less ric: there is a significant negative correlation (r ϭϪ0.53; p Ͻ frequent than peaks at other contrasts, indicating that cells fa-

0.001) between Sbright and Sdark. This suggests that V4 neurons do vored peaks at the extreme contrasts, as is common throughout ϭ not evenly cover the space defined by Sbright and Sdark; some com- , or at Clum 0, which is the novel class of equilu- Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12405

0.75 Bright Cells Dark Cells ing luminance contrast. In particular, com- paring contrast cells to equiluminance cells, 0.5 n = 49 n = 27 the response to 0% contrast (red lines, bot- tom row) rises later than the other re- 0.25 125% sponses, despite the reversal of the relative 50% 0% amplitudes of the signals. It is interesting to -50% note that the peak times for the high con- 0 trast and equiluminance population re- 0.75 Contrast cells Equiluminance cells Flat cells sponses in V4 correspond remarkably well with the optimal stimulus duration for the 0.5 n = 51 n = 44 n = 21 detection of luminance (55 ms) and chrominance (140 ms) flashes, respectively 0.25 (Chaparro et al., 1993). Normalized population response Since responses of equiluminance cells 0 0200400 02004000 200 400 to color stimuli are suppressed in the pres- Time relative to stimulus onset (ms) ence of high luminance contrast, we con- sidered the possibility that high color Figure 6. Time course of the population response for each cell class. Population PSTHs show the average normalized response (see contrast, just like high luminance con- Materials and Methods) relative to stimulus onset for bright, dark, contrast, equiluminance, and flat cells. Number of cells contributing to trast, also suppresses the responses of eachhistogramisindicatedbyn.Becausefewcellswerecategorizedaspeakcells(nϭ5),thecorrespondingPSTHsarenotshown.Stimulus these cells. We constructed population durationwas300msexceptforafewcells(3–6ineachcategory;500ms).Thelinecolorrepresentsstimuluscontrastasperthelegend.The PSTHs with the chromaticities split into response to near equiluminant stimuli (red line) was always the latest response, regardless of whether it was the strongest or weakest three color contrast bands (Fig. 1A) (see response on average. As in the examples in Figure 2, bright, contrast, and dark cells responded strongest to high-contrast stimuli, while Materials and Methods). Figure 7 shows equiluminance cells responded best to 0% contrast. color contrast PSTHs for stimuli at 0% luminance contrast in shades of red. minance cells described here. Thus, equiluminance cells are al- Across the population of equiluminance cells, peak amplitude is most as prevalent as bright or contrast cells in area V4 but more quite similar for all three color contrast bands; specifically, peak prevalent than cells with peaks at intermediate positive contrasts. amplitude for high color contrast was not weaker than for the This does not argue against equiluminance cells being a part of a lower color contrasts. This suggests that the suppression of equi- continuous distribution of contrast response functions in V4, but luminance cell responses is specific to luminance contrast. Inter- it does suggest that they are an important part of that continuum estingly, however, color contrast, like luminance contrast, and could represent a specialization in V4 for the representation appears to modulate response latency: responses emerged earlier of equiluminant form (see Discussion). for stimuli at higher color contrast. The trend in latency as a To control for the unlikely possibility that equiluminance cells function of color contrast is not evident for stimuli at a higher prefer the actual background luminance (5.4 cd/m 2) rather than 0% luminance contrast (blue lines), presumably because the effect of luminance contrast, we studied three equiluminance cells at two luminance contrast swamped the more subtle changes due to additional background luminances (2.7 and 8.1 cd/m 2). All de- color contrast. This gives a coherent impression that contrast, tails of the experiment were identical except for the background whether luminance or color, is a fundamental driver of response luminance. Results for an example neuron at all three back- strength, as reflected by latency. Nevertheless, equiluminance ground luminances are shown in Figure 5B. The three curves cells must have specific circuitry to reverse the amplitude rela- peaked at different luminances depending on the background tionship, but not the time relationship. luminance: the peak position matched the background lumi- All of the results presented so far are based on photometric ϭ nance. This indicates that the neuronal response reflected a pref- equiluminance (i.e., Clum 0 as defined using a photometer and erence for the stimulus contrast rather than the luminance itself. not the behaviorally determined equiluminance points of the an- All three control neurons showed the same pattern of results. imal or that of any given cell). Because the equiluminance point Consistent with this result, if equiluminance cells preferred can vary between species, individual subjects, and even between the actual luminance rather than the 0% contrast, one would cells, we sampled the contrast axis more finely about photometric expect the distribution of preferred luminances to be evenly equiluminance for 13 equiluminance cells and verified that re- distributed between the four values tested, but this was not the sponses in the Ϫ5 to 5% contrast range were similarly high and case (see above). These analyses support the hypothesis that showed a gradual decline to baseline levels near Ϯ25% contrast luminance contrast, rather than luminance, was the relevant (Fig. 8). In every case, the chromatic contrast response function stimulus attribute. ϭ (red line) was unimodal with a peak close to Clum 0, and the Having found a class of cells that respond best at low lumi- best achromatic contrast response (black line) was always weaker nance contrasts, we examined whether these cells also turn on its than the best chromatic response near equiluminance. These re- head the usual latency relationship with luminance contrast: sults support the conclusion that equiluminance cells respond high-contrast signals typically produce responses that are not best to preferred chromaticities near equiluminance and that only stronger but that also tend to occur sooner. Figure 6 shows these responses cannot be attributed to selectivity for low achro- the population PSTHs for the different cell categories and the matic contrast stimuli. results match the corresponding single cell examples in Figure 2. Bright, dark, and contrast cell populations responded best to high-contrast stimuli, whereas equiluminance cells showed stron- Comparison with V1 ϭ gest responses at Clum 0. In terms of response latency, however, all To examine how our results compare with previous findings in cell classes followed a similar trend: latency decreased with increas- V1, we characterized our cells in terms of their color sensitivity 12406 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4

Equiluminance cells A Single cell examples 0.75 n = 44 high color Chromatic mid 30 20 contrast Achromatic low 20 0% Lum contrast 10 0.5 10 125% lum 0 contrast 0

0.25 15 40

Spikes/sec 10 30 Normalized response 5 20 0 0 10 0 200 400 Time relative to stimulus onset (ms) −50 −25 0 25 50 −50 −25 0 25 50 Clum (%) Figure 7. Population PSTHs as a function of color contrast. Stimuli were divided into three B groups based on color contrast: chromaticities 2–7 (low color contrast band), 8–13 (mid color contrast band), and 14–25 (high color contrast band). Population PSTHs for equiluminance cells based on these three groups of stimuli are shown for stimuli at 0% luminance contrast 1.5 Population (shadesofred)and125%luminancecontrast(shadesofblue).Thex-andy-axesareasinFigure 6.At0%luminancecontrast,peakamplitudesaresimilarforthethreecolorcontrasts,suggest- ing that responses of equiluminance cells to high color contrast are not also suppressed. Re- 1 sponses emerged earlier for higher color contrasts for stimuli at 0% luminance contrast but not for stimuli at 125% luminance contrast.

0.5 (Johnson et al., 2001) and in terms of a linear cone weight model (Lennie et al., 1990). For each cell, we calculated a color sensitiv- ϭ 0 ity index, SI Re/Ra, where Re is the best response (in excess of −50 −25 0 25 50 Normalized population response baseline) in the equiluminant plane and Ra is the best response to Clum (%) achromatic stimuli. The distribution of SI is shown in Figure 9A for all cells, arranged by the cell categories defined in Figure 4. Figure 8. Finer sampling of the contrast response function. A, Each panel shows the re- The SI values for bright, dark, and contrast cells were centered sponses of an equiluminance cell to a preferred chromaticity (red line) and the achromatic ϭ Ϯ Ϯ Ϯ Ϯ near 1.0 and most ranged between 0.5 and 2.0, indicating that the stimulus(blackline)presentedatClum 0, 2.5, 5, 25,and 50%luminancecontrasts. best achromatic and equiluminant responses differed by no more Mean responses (across 20 repetitions) are plotted against luminance contrast (x-axis). Error than a factor of 2. This is similar to the range of values exhibited bars represent SEM. In four representative single cells, chromatic responses tended to be high by V1 color luminance cells described by Johnson et al. (2001). near 0% contrast and gradually declined for higher contrasts (25–50%). Responses to achro- Most equiluminance cells, however, had SI Ͼ 2.0 (i.e., the re- matic contrasts were typically bimodal with weakest responses at zero and high contrasts. Responses to the best achromatic contrast were weaker than to chromatic stimuli near equilu- sponse to the best equiluminant stimulus was at least twice as minance.B,Normalizedpopulationresponseacrossall13cellsonwhichthefinersamplingwas large as the response to the best achromatic stimulus). A small conducted. The responses of each cell were normalized by the response to the chromatic stim- proportion (ϳ11%) of V1 cells also exhibit indices Ͼ2.0 (color ϭ ulusat0%luminancecontrast.Thus,atClum 0,yfortheredlineequals1bydefinition.Across cells; Johnson et al., 2001). Compared with V1, far fewer neurons the population, chromatic contrast response functions were quite narrow: responses decline to had indices Ͻ0.5, where the best achromatic response was at least 50% of the peak at Ϯ25% contrast. For this range of contrasts, responses to achromatic con- twice as large as the best equiluminant response (luminance cells; trasts were much weaker. Note that the x-axes here range from Ϫ50 to ϩ50, unlike in Figures Johnson et al., 2001). To assess whether this was simply due to the 2 and 4, in which they range from Ϫ50 to ϩ125. imbalance in our comparison—Re was the maximum of 4 num- bers, while Ra was the maximum of 24 numbers—we computed The observation that equiluminance cells occupy the right tail an SI distribution based on the maximum of 3 numbers for each. of the SI distribution in Figure 9A is expected by definition be- We redefined Ra as the maximum response to the three nonzero cause cells responding best to stimuli in the equiluminant plane achromatic contrasts and Re as the maximum response to the and weaker to higher contrasts were classified as equiluminance most saturated red, green, and blue in our stimulus set (Fig. 1A, cells. However, that most contrast cells and a large fraction of symbols 16, 24, and 20). In this case, 17% of V4 neurons exhibited bright and dark cells do not occupy the space defined by SI Ͼ 2.0 values of SI Ͻ 0.5 (up from 5% in Fig. 9A, as expected), but this does not follow from our categorization scheme. A large fraction was still substantially less than the proportion observed in V1 of these cells are clustered close to SI ϭ 1.0 (i.e., Re Ϸ Ra). Such (60%; Johnson et al., 2001). Further experiments are needed to neurons cannot have balanced cone opponency if their responses determine whether this difference between cortical areas reflects a arise from linear weighting of cone excitation because balanced specialization of function, or is simply attributable to the fact that opponency in the linear model implies Re Ͼ Ra. For instance, in ϭϪ we tested more color directions in the equiluminant plane than Equation 1, if wL wM, stimuli in the equiluminant plane with Johnson et al. (2001). For the 13 equiluminance cells with finely CM and CL of opposite sign will evoke greater responses than sampled chromatic and achromatic contrast response functions achromatic stimuli (CM and CL of the same sign). (Fig. 8), SI values similarly stood apart from the other groups of To determine whether a linear combination of cone excitation cells when Ra and Re were based on the finely sampled contrast signals can explain V4 responses and to assess the sign and mag- response functions. nitude of the corresponding cone weights, for each neuron, we Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12407

fit for the dark cell in Figure 2B (88% explained variance). In this A ϭϪ ϭϪ Color Sensitivity Index case, a large negative M-cone weight (WL 0.02; WM 0.9; ϭϪ 200 WS 0.08) successfully predicted strong responses to stimuli darker than the background and close to the red end of the color space. In contrast, the models considered here provided poor fits for the contrast and equiluminance cell responses in Figure 2, C and D, capturing only 16 and 7% of the response variance, re- spectively. For the cell in Figure 2D, the best fitting model pre- 100 dicted a high baseline (7.4 spikes/s) and negative weights for all ϭϪ ϭϪ ϭϪ three cones (WL 0.08; WM 0.85; WS 0.07). Thus, ϭϪ predicted responses were highest for Clum 50%, intermediate

Cell Number ϭ ϭ for Clum 0 and lowest for Clum 125%. The linear cone weight model is particularly poor at explaining responses to equilumi- nance cells because, for these cells, the highest responses are at intermediate values for cone contrast. This produces a bell- 0 shaped relationship between the independent variables (cone 0.01 0.1 1 10 100 contrasts) and the dependent variable (response) that cannot be Sensitivity Index (SI) successfully modeled by a linear weighting of cone contrasts. Re- B Goodness of fit sults across the population are illustrated in Figure 9B. Linear 20 Bright Dark cone excitation-based models, with or without rectification, pro- 15 vided a good fit for many bright and dark cells in our population, but fits were poor for a large fraction of the equiluminance cells. 10 The mean percentage explained variance for the bright, dark, contrast, and equiluminance cells were 35.7, 50.1, 32.0, and 17.9, 5 respectively. In terms of opponency, among the cells with signif- 0 icant linear fits (bright, 48 of 49; dark, 25 of 27; contrast, 51 of 51; 20 equiluminance, 28 of 44), 60% of bright cells, 48% of dark, 31% Contrast Equiluminance of contrast, and 75% of equiluminance cells showed varying de- 15 grees of opponency in cone weights. Median cone weight ratio Number of cells Ϫ (WL/WM) was 0.2 for bright and dark cells, 0.98 for contrast 10 cells, and Ϫ0.7 for equiluminance cells that had significant linear 5 fits. These results suggest that there was a trend toward weak opponency among bright and dark cells, no opponency among 0 contrast cells, and balanced opponency among equiluminance 0 0.3 0.6 0.9 0 0.3 0.6 0.9 cells, consistent with the observations based on the spread of SI Percent explained variance indices above. In summary, these comparisons suggest that a larger fraction Figure 9. Color sensitivity indices of V4 neurons and the goodness of fit of linear cone exci- of our V4 population displayed higher SI values than do V1 cells, tation models with or without rectification. A, Color sensitivity index, SI, is the ratio of the maximum response among equiluminant stimuli, Re, to the maximum response among achro- largely as a result of the existence of equiluminance cells, and that maticstimuli,Ra.EachpointshowsSIforonecell.ColorrepresentstheclassificationfromFigure while the responses of some V4 cells were well described by a 4.Thedottedlinesidentifytherangeofsensitivityindices(0.5ϽSIϽ2.0)thatwereclassified linear cone model, many were not. The latter findings are consis- as color luminance cells by Johnson et al. (2001). Sensitivity indices of the vast majority of V4 tent with reports from V1 and V2 that many color-selective neu- neurons, except most equiluminance cells, fell in this range. Most equiluminance cells, how- rons exhibit nonlinear combination of cone signals (Kiper et al., ever,hadsensitivityindicesϾ2.0.For14cells(3bright,1dark,9equiluminance,and1flatcell), 1997; Hanazawa et al., 2000; Wachtler et al., 2003). Further stud- Ra was zero; these cells are shown at the right extreme (note broken x-axis). B, Histogram of ies are needed to develop more appropriate nonlinear models to goodness of fit values for bright (top left), dark (top right), contrast (bottom left), and equilu- describe and determine how V4 color responses arise, especially minance (bottom right) cells. For each cell, we fit the neuronal responses with linear cone for equiluminance cells. excitation models with or without rectification (see Materials and Methods). Goodness of fit (x-axis)isgivenbythepercentageofvarianceexplainedbythebestfittingmodel.Somebright, dark, and contrast cells were well described by a rectified linear model, but for most equilumi- Coarse color and shape preferences of equiluminance cells nance cells, linear cone excitation-based models provided a poor fit of the data (see Results). To determine whether equiluminance cells differed from other Mean goodness of fit for bright, dark, contrast, and equiluminance cells were 35.7, 50.1, 32.0, V4 cells along stimulus dimensions other than luminance con- and 17.9, respectively. trast, we compared cell categories in terms of coarse color (Figs. 10, 11) and shape preferences (Fig. 12). More detailed character- ization of the chromatic and shape preferences of V4 cells has estimated the relative cone weights, WL, WM, and WS, that best been addressed in previous studies (Desimone and Schein, 1987; predicted the observed responses. To allow for nonlinearities, we Schein and Desimone, 1990; Kobatake and Tanaka, 1994; Pasu- also considered models with half- and full-wave rectification (see pathy and Connor, 2001; Kusunoki et al., 2006; Conway et al., Materials and Methods). For some neurons, such as the bright cell in 2007; Kotake et al., 2009); we simply ask whether the classes of Figure 2A, a half-wave rectified linear model provided an excellent cells defined here on the basis of luminance contrast response fit to the observed responses, capturing 80% of the response vari- functions can be differentiated by their chromatic and shape ance. For this neuron, the L and M weights were nonopponent and preferences. For every cell, we characterized the numbers of col- ϭ ϭ ϭ the S cone weights were small (WL 0.37; WM 0.62; WS ors that evoked responses greater than baseline (Ncol_sig) and Ϫ 0.01). A half-wave rectified linear model also provided a good greater than one-half of the maximum response (Ncol_hmax). 12408 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4

AB categories, preferred colors spanned the tested range (Fig. 10C– H), consistent with previous results (Komatsu et al., 1992). Fi- 30 nally, we found no significant variation in peak firing rates, with 20 or without baseline subtraction, across the cell classes (random- 20 ization ANOVA, p Ͻ 0.05). In terms of hue tuning, equiluminance cells can be broadly col_sig 10 classified into three groups: (1) cells that show strong, unimodal N

10 col_hmax

N tuning; (2) cells that show complex bimodal or multimodal tun- ing; and (3) cells that do not show statistically significant tuning. 0 0 Figure 11 illustrates the hue response surfaces and tuning curves BDCEFP BDCEFP for six representative equiluminance cell examples. Hue response Cell classes Cell classes surfaces in the CIE color space (Fig. 11, images) were constructed 0.6 by linear interpolation of normalized responses to stimuli at 0% C BrightD DarkE Contrast Number of cells 4 luminance contrast. To assess the influence of color saturation on hue tuning, hue tuning curves were constructed by plotting re-

0.3 sponse as a function of color direction relative to the achromatic 2 point for each of the three bands of color contrasts tested (Fig. 1A) (see Materials and Methods). Figure 11A–C shows examples that exhibited narrow, unimodal tuning in the CIE color space. 0 0 0.6 The neuron in Figure 11A responded preferentially to green col- Equiluminance Flat Peak CIE y F G H ors at the highest color contrast (black line). Responses to mid and low color contrasts were weaker, but the peak position is similar. Responses at the highest color contrast showed statisti- 0.3 cally significant modulation as a function of color direction (Ray- leigh test of circular uniformity, p Ͻ 0.05, Bonferroni corrected). Strength of unimodal tuning was quantified as the resultant vec- 0 tor length (see Materials and Methods), which ranges from 0 (no 00.30.60 0.3 0.6 0 0.3 0.6 CIE x unimodal tuning) to 1 (strong unimodal tuning). For the neuron in Figure 11A, the resultant vector length based on the responses Figure 10. Color preferences across the V4 population. A, Mean and SD across cells for the to high color contrast stimuli was 0.72. Figure 11, B and C, shows number of colors (of 25) that elicited significant responses different from baseline for at least two other examples of equiluminance cells with strong unimodal oneofthefourluminancecontraststested;cellsaregroupedintocategories(perFig.4)labeled tuning. Both neurons showed significant modulation of re- herebythefirstletterofthecategoryname:bright(nϭ49),dark(nϭ27),contrast(nϭ51), sponses as a function of color direction at one or more color equiluminance(nϭ44),flat(nϭ21),andpeakcells(nϭ5).B,Numberofcolorsthatelicited contrasts and the resultant vector length at the preferred color more than half-maximum responses at the preferred contrast. C–H, Distribution of preferred contrast was 0.5 and 0.74, respectively. colors that elicited the maximum response for neurons grouped by category. The x- and y-axes Unlike the preceding examples, the hue tuning curves of ex- correspondtoCIExycoordinates.Grayscaleindicatesthenumberofcellsthatrespondedbestto amples D and E are more erratic and complex, but statistical tests the corresponding color. In several cases, rectangles are white and not visible because no cell indicate that these responses also showed significant modulation had a peak response at the corresponding color. as a function of color direction at the mid (dark gray) and low (pale gray) color contrasts, respectively. The response surface of

Across cells, Ncol_sig ranged between 2 and 25 (Fig. 10A), and the example D suggests that this neuron had two peaks in opposite category averages ranged between 11 (for peak cells) and 22 (for color directions from the achromatic point, and this is reflected contrast cells). For equiluminance cells, Ncol_sig was on average 17 in the hue tuning curve for mid contrasts (dark gray); the resul- colors (SD, 7.2), indicating that every equiluminance cell was tant vector length was 0.2 consistent with the non-unimodal tun- driven by many colors and that the contrast function character- ing curves. Example E, however, showed broad color tuning, izations were not based on only a few colors. Average values for responding to most equiluminant colors and the resultant vector

Ncol_hmax ranged between 6 and 14 (Fig. 10B), and for equilumi- length was again low (0.17). Finally, example F showed an annu- nance cells was 8.8 (SD, 5). Thus, near equiluminance, roughly lar response pattern in color space: responses were strong for all one-third of the colors evoked strong responses from equilumi- color directions with the exclusion of a small region centered on nance cells, suggesting that these cells were broadly color selec- the achromatic point and the hue tuning in this case was not tive. Both measures, Ncol_sig and Ncol_hmax, showed a significant significantly different from uniform tuning. dependence on the cell class (randomization one-way ANOVA, Across our population, 18% of equiluminance cells showed p Ͻ 0.01), but the values observed for equiluminance cells were strong unimodal tuning like the examples A, B, and C (resultant within the range of values observed across all classes of cells de- vector length, Ͼ0.4) and 30% showed bimodal or multimodal scribed here. To take into account the possibility that Ncol_hmax tuning or broad and weaker unimodal tuning (as in examples D may be underestimated for equiluminance cells because low and E). The remaining 52% of neurons (like example F) showed color contrast stimuli (at 0% luminance contrast) may be inef- no significant response modulation as a function of hue. These fective at driving these cells, we quantified Ncol_hmax_highcontrast proportions were similar to those observed for bright cells (16 based on only the 12 high color contrast stimuli (chromaticities and 27% for unimodal and multimodal, respectively), but the 14–25; Fig. 1A) and our results were consistent with those re- proportion of strongly unimodal cells was smaller among dark ported above [i.e., Ncol_hmax_highcontrast showed significant depen- (8%) and contrast cells (2%). dence on cell class, but values exhibited by equiluminance cells Equiluminance cells are similar to the general V4 population (4 Ϯ 2.5) were within the range across all cells (2–7)]. For all cell in terms of the variety and complexity of their hue-tuning prop- Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12409

10 A Color Contrast D High 40 Mid 5 Low 20

0 0

24 B 20 E

16 10

8 0 Fring rate (spikes/s) Fring rate (spikes/s) 1 C 30 F 30

20 20 0.5 10 10

0 0 0

Normalized response 0 90 180 270 360 0 90 180 270 360 Color direction (°) Color direction (°)

Figure11. Huetuningofsixexampleequiluminancecells.A–F,Colorimagesshowthenormalized,linearlyinterpolatedresponsesurfacebasedonresponsestostimuliat0%luminancecontrast. Color scale runs from blue to black to red representing low to high responses in the CIE space. Corresponding line plots show raw responses as a function of color direction in the CIE chromaticity diagram.Colordirection(x-axis)wasmeasuredrelativetotheachromaticpoint,0°istotheright,andanglesincreasecounterclockwise.Thered,green,andbluearrowsmarkthecorrespondingcolor directions. The pale gray, dark gray, and black lines represent responses to stimuli of low, mid, and high color contrasts, respectively. Error bars show SEM. The horizontal dashed line indicates the baselineresponse.ExamplesA–Cshowunimodaltuningforhueatoneormoresaturations.ExamplesDandEalsoshowstatisticallysignificantmodulationofresponseasafunctionofcolordirection but lack unimodal hue tuning. Example F (same cell as in Fig. 3D) did not differ significantly from uniform tuning. See Results for details.

A B al., 2007). In our data set, only 10% of all cells showed significant pairwise correlations. This discrepancy might arise because we 1 1 tested four luminances, and thus had six pairings, whereas they 0.75 0.75 tested three luminances and had three pairings. Lack of signifi- cant correlation between responses at different luminances can 0.5 0.5 result if (1) strong modulation by luminance contrast results in lack of responses at one or more of the tested luminances (for shapes_sig 0.25 0.25 shapes_hmax F

F example, see Fig. 3D), or (2) chromatic tuning is dependent on 0 0 luminance tuning such that the neuron prefers different chroma- BDCEFP BDCEFP ticities at different luminances. The SVD analysis (see Materials Cell classes Cell classes and Methods) directly tested which of these two reasons underlie Figure 12. Shape preferences across the V4 population. A, Mean and SD across cells for the the lack of correlation across some luminance pairings in our fraction of shapes that evoked responses significantly different from baseline. B, Mean and SD data. Because all except one of the neurons in our data set exhib- for the fraction of shapes that evoked more than half-maximum responses. The x-axis shows ited strong separability between the influence of chromaticity and first letter of cell class names, as in Figure 10A. luminance on neuronal responses, the lack of significant correla- tion across some luminance pairings is attributable to the former erties. Cells with narrow tuning for hue and saturation, like the rather than the latter reason above. In summary, while colored cells in Figure 11A–C, have been previously reported in areas V1, stimuli are necessary to evoke strong responses from equilumi- V4, and IT cortex (Hanazawa et al., 2000; Kusunoki et al., 2006; nance cells, strong hue tuning is not a defining feature of these Conway et al., 2007; Kotake et al., 2009). While equiluminance cells. Some equiluminance cells show strong hue tuning, but cells have not been previously reported, responses of an example many cells show weak or no hue tuning consistent with the gen- cell from a glob region in the study by Conway et al. (2007), their eral V4 population. Figure 6A, panel 2, are consistent with those of a strongly hue- To investigate whether stimulus shape modulates the re- tuned equiluminance cell. Cells with tuning for multiple color sponses of equiluminance cells and to assess differences in shape directions or broad unimodal tuning have also been previously selectivity across cell classes, we examined the responses of a sub- described in V4 (Kusunoki et al., 2006; Conway et al., 2007; Ko- set of 118 neurons that had been studied with a set of 10–29 take et al., 2009). One previous V4 study investigated the consis- shapes (Fig. 1B), each presented at multiple orientations. All tency of color tuning across luminances by quantifying the classes of neurons showed similar levels of shape selectivity. On correlation coefficient between responses to color stimuli at dif- average, the fraction of stimuli that evoked significant responses ferent luminances and found that 40% of interglob cells showed (Fshape_sig) ranged between 0.32 and 0.64; the fraction that evoked significant pairwise correlations for all comparisons (Conway et greater than half-maximum responses (Fshape_hmax) ranged be- 12410 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4 tween 0.18 and 0.47. There were no systematic differences in contrast, luminance contrast, or both (Shapley and Hawken,

Fshape_sig and Fshape_hmax across cell groups (randomization 2002). The bright, dark, contrast, flat, and peak cells in our pop- ANOVA). Among equiluminance cells, on average, Fshape_sig was ulation have response properties similar to color luminance cells 0.61 and Fshape_hmax was 0.29. Thus, the different shapes used in in V1—these cells show selective responses to complex shapes this study modulated the responses of equiluminance cells and and respond equally well to equiluminant and achromatic stim- other V4 cells similarly. While this analysis does not characterize uli. Thus, the response properties of these neurons could be de- shape-tuning properties of these cells, it does suggest that the rived from color luminance cells in V1. responses of equiluminance cells are modulated by, and could A small fraction, ϳ10%, of V1 neurons respond stronger to encode, information about visual shape. Again, maximum firing equiluminant stimuli than to luminance stimuli (Lennie et al., rate on the shape tuning test did not show a significant depen- 1990; Johnson et al., 2001). But because these color-sensitive neu- dence on cell category. rons typically have low-pass spatial frequency tuning curves and respond well to extended chromatic surfaces, the properties of V4 Discussion equiluminance cells are unlikely to be derived solely from these Equiluminance cells appear to represent a major physiological V1 responses. Responses of equiluminance cells to their preferred class of neuron in V4 that are similar to the general V4 population colors weaken when those chromaticities are presented at higher in terms of shape and color selectivity but quite unlike other luminance contrasts. Therefore, linear models based on cone ex- established classes of cells anywhere in visual cortex in terms of citation or contrast fail to explain their responses (see Results). their selectivity for luminance contrast. Specifically, equilumi- Response properties of equiluminance cells could arise by com- nance cells have strong responses near equiluminance and bining excitatory chromatic inputs with inhibitory inputs that weaker responses at higher contrasts. Strong responses to equilu- elevate with increasing contrast. For example, equiluminance minant stimuli are not a surprising finding: many neurons in V1 cells could receive an especially strong inhibition akin to contrast and V2 respond strongly to equiluminant stimuli (Gouras and normalization. To test for this, we superimposed high-contrast Kruger, 1979; Hubel and Livingstone, 1990; Lennie et al., 1990; achromatic gratings of varying spatial frequencies on preferred Kiper et al., 1997). The novel and surprising finding here is that color equiluminant stimuli and found that this did not reproduce high luminance contrasts can suppress chromatic responses to the suppression we observed by increasing the luminance con- levels substantially lower than those observed near equilumi- trast of the shape (data not shown). This suggests that pattern- nance. To our knowledge, no previous study has characterized specific suppressive mechanisms are involved in the construction responses in visual cortex that monotonically decrease with in- of equiluminance cells, perhaps requiring coextensive luminance creasing luminance contrast within the classical receptive field. and color gradients. It is implausible that our results are due to Perhaps some equiluminance cells exist in V1 and V2; however, luminance artifacts generated by chromatic aberration, because similar experiments, which modulated luminance contrast of equiluminance cells show unimodal response curves and do not color stimuli, have been conducted in V1 (Gouras and Kruger, respond to high luminance contrasts. In other words, being off by 1979; Thorell et al., 1984; Hubel and Livingstone, 1990) and V2 a modest amount from ideal equiluminance for each cell would (Kiper et al., 1997), and no such neurons have been described. not change the overall interpretation of our data. Past studies Hubel and Livingstone (1990) have documented that a few neu- have demonstrated that V4 color responses can be influenced by rons in layer 2/3 of V1 show a shallow dip or no dip at all at task requirements and top-down signals including attention equiluminance, analogous to flat cells described here. Apart from (Motter, 1994; Ogawa and Komatsu, 2004; Bichot et al., 2005). that, in analogous studies, V1 and V2 responses increased with Such influences cannot explain the different categories of neu- the incremental addition of luminance to chromatic stimuli rons reported here because these animals were not trained on any (Gouras and Kruger, 1979; Thorell et al., 1984; Kiper et al., 1997). task other than passive fixation and the proportions of different While it is possible that V4 inherits equiluminance selectivity categories of neurons were similar for the two animals. from other areas, given the absence of such cells in previous de- Several psychophysical studies have investigated the extent tailed studies of color selectivity in V1 and V2, it seems more of segregation and overlap in the processing of chromatic and likely that the response properties of V4 equiluminance cells luminance stimuli. Evidence from experiments using achro- emerge de novo in V4. matic luminance and equiluminant color stimuli (but not It has long been hypothesized that double-opponent cells, their combination) support the existence of a color-based which show chromatic and spatial opponency, play an important shape analysis system that is independent of luminance infor- role in encoding simultaneous color contrast and chromatically mation. Because simultaneous and opposite spatial frequency defined form (Daw, 1967; Livingstone and Hubel, 1984; Michael, shifts (Favreau and Cavanagh, 1981) and tilt aftereffects (Flana- 1985; Shapley and Hawken, 2002; Conway, 2009). The precise gan et al., 1990) were obtained in adaptation experiments with definition and RF structure of double opponent cells has been equiluminant-chromatic and achromatic gratings, it has been ar- widely debated: while one view is that they compose ϳ5–10% of gued that a channel exists that codes spatial frequency and orien- V1 and have coarse receptive field structure (Conway, 2001; Con- tation solely based on chromatic information. Results from way and Livingstone, 2006), others argue that they compose adaptation experiments measuring detection thresholds (Bradley ϳ30% of V1 (color luminance cells; Johnson et al., 2001) and that et al., 1988; Murasugi and Cavanagh, 1988) are consistent with they are well equipped for form vision (Shapley and Hawken, this hypothesis, and results from visual search experiments (Ca- 2002). Regardless of how their RFs arise and whether they fit the vanagh et al., 1990) indicate that orientation and size are basic precise definition of double opponency, these color luminance coding dimensions for equiluminous colors. These results point cells respond well to achromatic luminance and equiluminant to the existence of a chrominance channel that analyzes shape stimuli (Lennie et al., 1990; Johnson et al., 2001; Wachtler et al., information solely based on chromatic signals (Cavanagh, 1991). 2003). They are tuned to stimulus orientation and high spatial More extensive investigation along multiple directions in the frequencies (Johnson et al., 2001) making them well suited to chrominance by luminance space supports the existence of mul- play a role in encoding object boundaries defined by chromatic tiple color mechanisms, not just a second chrominance-based Bushnell et al. • Equiluminance Cells in V4 J. Neurosci., August 31, 2011 • 31(35):12398–12412 • 12411 mechanism (Gegenfurtner and Kiper, 1992; Hansen and Gegen- Chaparro A, Stromeyer CF 3rd, Huang EP, Kronauer RE, Eskew RT Jr (1993) furtner, 2006). Results from these studies, which test contrast Colour is what the eye sees best. Nature 361:348–350. detection and image segmentation in the presence of noise, reveal Cole GR, Hine T (1992) Computation of cone contrasts for . Behav Res Methods Instrum Comput 24:22–27. that behavioral performance is influenced by noise over a narrow ϫ Conway BR (2001) Spatial structure of cone inputs to color cells in alert range centered on the stimulus direction in the chrominance macaque primary visual cortex (V-1). J Neurosci 21:2768–2783. luminance space. This supports the hypothesis of multiple color Conway BR (2009) Color vision, cones, and color-coding in the cortex. mechanisms with peaks at different chrominance ϫ luminance Neuroscientist 15:274–290. combinations; the narrow tuning widths suggest interaction be- Conway BR, Livingstone MS (2006) Spatial and temporal properties of cone tween linear mechanisms. signals in alert macaque primary visual cortex. J Neurosci 26:10826- Equiluminance cells in V4 have properties that correspond 10846. Conway BR, Moeller S, Tsao DY (2007) Specialized color modules in ma- well with the requirements of the shape-encoding chrominance caque extrastriate cortex. Neuron 56:560–573. channel proposed above: (1) they are the first set of visual neu- Daw NW (1967) Goldfish : organization for simultaneous color con- rons that have been shown to be selective for low luminance trast. Science 158:942–944. contrasts, where shape information is largely based on color con- Depireux DA, Simon JZ, Klein DJ, Shamma SA (2001) Spectro-temporal trast; (2) they are shape selective; and (3) they form a sizable response field characterization with dynamic ripples in ferret primary subpopulation of V4. The narrow tuning of equiluminance cells auditory cortex. J Neurophysiol 85:1220–1234. along the luminance contrast dimension (Fig. 8) is qualitatively Desimone R, Schein SJ (1987) Visual properties of neurons in area V4 of the macaque: sensitivity to stimulus form. J Neurophysiol 57:835–868. consistent with the narrow tuning width observed in the psycho- De Valois RL (1960) Color vision mechanisms in the monkey. J Gen Physiol physical experiments described above, thus lending physiological 43 [Suppl]:115–128. support for the proposal of multiple color mechanisms. Because Favreau OE, Cavanagh P (1981) Color and luminance: independent fre- equiluminant boundaries are more common than previously quency shifts. Science 212:831–832. thought and they provide a statistically independent source of Flanagan P, Cavanagh P, Favreau OE (1990) Independent orientation- shape information in natural images (Hansen and Gegenfurtner, selective mechanisms for the cardinal directions of colour space. Vision 2009), form information carried by equiluminance cells may play Res 30:769–778. Gawne TJ, Kjaer TW, Richmond BJ (1996) Latency: another potential code an important role in shape perception. for feature binding in striate cortex. J Neurophysiol 76:1356–1360. In conclusion, our results suggest that there is a particular Gegenfurtner KR, Kiper DC (1992) Contrast detection in luminance and class of cells in V4 that is specialized to represent and process chromatic noise. J Opt Soc Am A 9:1880–1888. near-equiluminant visual scenes, in which the information is Gegenfurtner KR, Kiper DC, Levitt JB (1997) Functional properties of neu- largely based on chromatic contrast. These equiluminance cells rons in macaque area V3. J Neurophysiol 77:1906–1923. are well suited to play a role in perception near equiluminance Gouras P, Kru¨ger J (1979) Responses of cells in foveal visual cortex of the monkey to pure color contrast. J Neurophysiol 42:850–860. and could be crucial for widely cited color tasks of foraging for Hanazawa A, Komatsu H, Murakami I (2000) Neural selectivity for hue and ripe fruit among green foliage and defeating luminance-matched saturation of colour in the primary visual cortex of the monkey. Eur camouflage. Our findings suggest that the representation of J Neurosci 12:1753–1763. scenes based on chromatic signals may be a specialization in V4 Hansen T, Gegenfurtner KR (2006) Higher level chromatic mechanisms for and supports the hypothesis that form at equiluminance and image segmentation. J Vis 6:239–259. form at higher contrasts are processed in separate channels (Ca- Hansen T, Gegenfurtner KR (2007) Higher order color mechanisms for im- vanagh, 1991; Gegenfurtner and Kiper, 1992; Hansen and Gegen- age segmentation. In: Brain, vision and artificial intelligence LNCS (Mele F, Ramella G, Santillo S, Ventriglia F, eds), pp 72–83. Heidelberg: furtner, 2006). Finally, these results motivate a host of questions, Springer. including how this selectivity for equiluminance is generated, and Hansen T, Gegenfurtner KR (2009) Independence of color and luminance how the signals studied here are used within and beyond V4, edges in natural scenes. Vis Neurosci 26:35–49. which are likely to inspire further experiments and fuel heated Hubel DH, Livingstone MS (1990) Color and contrast sensitivity in the lat- debates that have long marked the study of color vision (De eral geniculate body and primary visual cortex of the macaque monkey. Valois, 1960). J Neurosci 10:2223–2237. Johnson EN, Hawken MJ, Shapley R (2001) The spatial transformation of color in the primary visual cortex of the macaque monkey. Nat Neurosci References 4:409–416. Albrecht DG (1995) Visual cortex neurons in monkey and cat: effect of Kiper DC, Fenstemaker SB, Gegenfurtner KR (1997) Chromatic properties contrast on the spatial and temporal phase transfer functions. Vis Neuro- of neurons in macaque area V2. Vis Neurosci 14:1061–1072. sci 12:1191–1210. Kobatake E, Tanaka K (1994) Neuronal selectivities to complex object fea- Bichot NP, Rossi AF, Desimone R (2005) Parallel and serial neural mecha- tures in the ventral visual pathway of the macaque . J Neu- nisms for visual search in macaque area V4. Science 308:529–534. rophysiol 71:856–867. Bradley A, Switkes E, De Valois K (1988) Orientation and spatial frequency Komatsu H, Ideura Y, Kaji S, Yamane S (1992) Color selectivity of neurons selectivity of adaptation to color and luminance gratings. Vision Res in the inferior temporal cortex of the awake macaque monkey. J Neurosci 28:841–856. 12:408–424. Bushnell BN, Harding PJ, Kosai Y, Pasupathy A (2011) Partial occlusion Kotake Y, Morimoto H, Okazaki Y, Fujita I, Tamura H (2009) Organization modulates contour-based shape encoding in primate area V4. J Neurosci of color-selective neurons in macaque visual area V4. J Neurophysiol 31:4012–4024. 102:15–27. Carandini M, Heeger DJ, Movshon JA (1997) Linearity and normalization Kusunoki M, Moutoussis K, Zeki S (2006) Effect of background colors on in simple cells of the macaque primary visual cortex. J Neurosci the tuning of color-selective cells in monkey area V4. J Neurophysiol 17:8621–8644. 95:3047–3059. Cavanagh P (1991) Vision at equiluminance. In: Vision and visual dysfunc- Lee J, Williford T, Maunsell JH (2007) Spatial attention and the latency of tion: limits of vision (Kulikowski JJ, Murray IJ, Walsh V, eds), pp 234– neuronal responses in macaque area V4. J Neurosci 27:9632–9637. 250. Boca-Raton, FL: CRC. Lennie P, Krauskopf J, Sclar G (1990) Chromatic mechanisms in striate Cavanagh P, Arguin M, Treisman A (1990) Effect of surface medium on cortex of macaque. J Neurosci 10:649–669. visual search for orientation and size features. J Exp Psychol Hum Percept Livingstone MS, Hubel DH (1984) Anatomy and physiology of a color sys- Perform 16:479–491. tem in the primate visual cortex. J Neurosci 4:309–356. 12412 • J. Neurosci., August 31, 2011 • 31(35):12398–12412 Bushnell et al. • Equiluminance Cells in V4

Maunsell JH, Ghose GM, Assad JA, McAdams CJ, Boudreau CE, Noerager BD Peng X, Van Essen DC (2005) Peaked encoding of relative luminance in (1999) Visual response latencies of magnocellular and parvocellular macaque areas V1 and V2. J Neurophysiol 93:1620–1632. LGN neurons in macaque monkeys. Vis Neurosci 16:1–14. Reynolds JH, Pasternak T, Desimone R (2000) Attention increases sensitiv- Mazer JA, Vinje WE, McDermott J, Schiller PH, Gallant JL (2002) Spatial ity of V4 neurons. Neuron 26:703–714. frequency and orientation tuning dynamics in area V1. Proc Natl Acad Sci Schein SJ, Desimone R (1990) Spectral properties of V4 neurons in the ma- U S A 99:1645–1650. caque. J Neurosci 10:3369–3389. Michael CR (1985) Laminar segregation of color cells in the monkey’s stri- Sclar G, Maunsell JH, Lennie P (1990) Coding of image contrast in central ate cortex. Vision Res 25:415–423. visual pathways of the macaque monkey. Vision Res 30:1–10. Motter BC (1994) Neural correlates of attentive selection for color or lumi- Shapley R, Hawken M (2002) Neural mechanisms for color perception in nance in extrastriate area V4. J Neurosci 14:2178–2189. the primary visual cortex. Curr Opin Neurobiol 12:426–432. Murasugi CM, Cavanagh P (1988) Anisotropy in the chromatic channel: a Smith VC, Pokorny J (1975) Spectral sensitivity of the foveal cone pigments horizontal-vertical effect. Spat Vis 3:281–291. Ogawa T, Komatsu H (2004) Target selection in area V4 during multidi- between 400 and 500 nm. Vision Res 15:161–171. mensional visual search task. J Neurosci 24:6371–6382. Thorell LG, De Valois RL, Albrecht DG (1984) Spatial mapping of monkey Pasupathy A, Connor CE (2001) Shape representation in area V4: position- V1 cells with pure color and luminance stimuli. Vision Res 24:751–769. specific tuning for boundary conformation. J Neurophysiol 86:2505- Wachtler T, Sejnowski TJ, Albright TD (2003) Representation of color stim- 2519. uli in awake macaque primary visual cortex. Neuron 37:681–691. Pen˜a JL, Konishi M (2001) Auditory spatial receptive fields created by mul- Zeki SM (1973) Colour coding in rhesus monkey prestriate cortex. Brain tiplication. Science 292:249–252. Res 53:422–427.